When Eugene Garfield developed his idea for the first citation indices his focus was on the collation of data relating to the most significant research documents in science. The assumption was made that scientists sought to communicate their research by means of articles published in scholarly journals, usually in the English language, and usually in major international or regional journals with a recognised impact. The justification for a selective rather than an inclusive approach to indexing was that the vast majority of significant scientific research findings are published in a limited number of journals and, therefore, the indexing of this set would facilitate quick access to the most significant research outputs needed in the development of new research (Garfield 1996). The selection policy of WoS still largely follows this procedure for its three flagship databases (see Testa 2016). While this presumed pattern of communication may remain largely true of the hard sciences, there are significant differences for the Social Sciences and the Humanities, whose disciplinary practices often include publication in other languages, journals and types of document. Hicks (2005), for example, stresses that social scientists publish their work not just in international journals but in books, national journals and 'non-scholarly literature' – documents aimed at a non-specialised audience outside of academia. However, estimating the exact balance of cited source forms is difficult and there is considerable variation between individual case studies. For example, research on citation practices in anthropology have indicated that between 40% (Choi 1988a) and 58% (Robinson and Posten 2005) of citations are to research published in monograph form.
Leydesdorff (2003, 90), examining outputs from 1989-99, showed that 79% of citations made from science journals indexed within WoS Science Citation Index relate to documents in journals listed within the Science Index. This figure dropped to 45% for Social Science journals. Mongeon and Paul-Hus (2016), making comparisons between journals included in the citation indices and those included in Ulrich's Global Directory of Periodicals (which covers more than 300,000 journals, newsletters, online publications and more, available online at www.ulrichsweb.com), showed that for the Social Sciences Ulrich's Directory lists 22,519 periodicals of which WoS has indexed 2,893; for the Arts and Humanities Ulrich's Directory lists 9559 periodicals of which 1,172 are indexed in WoS. Journal coverage in Scopus is better but still low (Social Sciences: Ulrich's lists 22,519/Scopus indexes 5682; Arts and Humanities: Ulrich's lists 9559/Scopus indexes 1781). However, this still assumes that research outputs from disciplines are published in a restricted set of same discipline-related journals and this is problematic when research is multidisciplinary. As discussed above, most archaeological research outputs appear in sources that have not been classified as 'archaeology' in the citation indices, so an exact assessment of journal coverage is impossible. It is fair to say, however, that the starting bibliographic data for this study will be missing data on research documents published in most local and some regional journals that have not been selected for indexing.
Beyond journals, archaeological research outputs are also published in the form of books, book chapters or conference proceedings. Examples of archaeological research outputs in these forms have been collected for this study, but the percentage that this sample represents is unknown. An approximate sense of the difference can be gained by comparing the numbers of books and conference proceedings details used in this study with the number listed by Scopus, which is believed to be a better index for such document types. In this study, WoS has provided data on 226 books, 1420 book chapters and 3038 conference proceedings: Scopus search results against the same search criteria and time frame include data on 1218 books, 3382 book chapters and 3126 conference proceedings. Larivière et al. (2006) have argued that humanities researchers publish more than 70% of their documents outside of journals. The actual publication practices of archaeologists, in particular their use of scientific publications, suggests that this general figure for the humanities may be too high for archaeology, but the potential absence of data on books and book chapters as citing documents is clear.
It is worth noting, finally, that the differences in publication practices between the Sciences, Social Sciences, Arts and Humanities has been recognised by the creators of the citation indices, and both are making efforts to capture the bibliometric data from these other forms of publication; for example, WoS has started a specific Book Citation Index to integrate books and their references (Web of Science 2016). The limitations in coverage of the citation indices discussed here, therefore, should reduce still more in the future, further aided by the pressure for researchers to publish their research where possible in major high-impact journals as a mechanism to enhance its evaluated quality within national systems of research evaluation.
Internet Archaeology is an open access journal. Except where otherwise noted, content from this work may be used under the terms of the Creative Commons Attribution 3.0 (CC BY) Unported licence, which permits unrestricted use, distribution, and reproduction in any medium, provided that attribution to the author(s), the title of the work, the Internet Archaeology journal and the relevant URL/DOI are given.